Method and Apparatus for Detecting Plane, and Robot Apparatus Having Apparatus for Detecting Plane

Abstract

In a plane detection apparatus, a plane detection unit (3) includes a line fitting block (4) to select a group of distance data points being in one plane from distance data forming an image and extract lines from the distance data point group, and a region growing block (5) to detect one or more planar regions existing in the image from a group of all lines included in the image and extracted by the line fitting block (4). The line fitting block (4) first draws a line D1 connecting end points of the distance data point group, searches a point of interest brk whose distance to the line L1 is largest, segments the data point group by the point of interest brk when the distance is larger than a predetermined threshold, and determines a line L2 by the least-squares method when the distance is smaller than the predetermined threshold. In case there exists a larger number of data points than a predetermine number on one side of the line L2, the data point group is determined to be in a zig-zag shape, the data point group is segmented by the point of interest brk. These operations are done repeatedly. Thus, a plurality of planes robust against noises is detected simultaneously and accurately from distance data including measurement noises.

Description

TECHNICAL FIELD [0001] The present invention generally relates to a method and apparatus for detecting plane, for detecting planes from three-dimensional distance data, and a robot apparatus having the apparatus for detecting plane installed therein, and more particularly to a method and apparatus for detecting plane, for detecting planes by the scan line grouping method, and a robot apparatus having the apparatus for detecting plane installed therein. [0002] This application claims the priority of the Japanese Patent Application No. 2004-077215 filed on Mar. 17, 2004, the entirety of which is incorporated by reference herein. BACKGROUND ART [0003] Various researches have been made of algorithms for detecting a plane from three-dimensional distance data. A plane thus detected can be used to have a locomotion robot apparatus, for example, bypass an obstacle or climb stairs. [0004] Generally, a plane is detected from distance information in the following steps: [0005] 1. Three-dimensi...

AI Technical Summary

Benefits of technology

[0035] In the above plane detection apparatus according to the present invention, the line fitting means extracts lines on the basis of the fact that three-dimensional distance data in one plane will be in one straight line. Since the distribution of the distance data points varies under the influence of noises, however, lines are fit adaptively to the distribution of distance data (called “adaptive line fitting”), which enables an accurate line fitting robust against noises. Also, determination, by the scan line grouping, of a plane from the plurality of lines thus fitted permits an accurate plane detection without taking, under the influence of noises or the like, a plurality of planes as one plane or a single plane as a plurality of planes.

[0036] Also, the line fitting means can extract lines accurately by extracting a group of distance data points estimated to be in one plane on the basis of the distance between the distance data points to re-estimate, based on the distribution of distance data points in the group of distance data points, whether the group of distance data points is in one plane, and extracting a group of distance data points once on the basis of the distance between the distance data points in a three-dimensional space to re-estimate, based on the data point distribution, whether the distance data points are in one plane.

[0041] Furthermore, the planar region grouping means can estimate, based on a difference between a lines-determined plane and reference plane, whether the lines lie coplanar with the reference plane. Namely, it is possible to extract a plane more accurately by judging, based on a root mean square residual of the plane equation, for example, whether the extraction has been influenced by noises or the lines are not coplanar with the reference plane.

[0055] The plane detection apparatus and method permit to accurately extract lines from even distance data with noises by fitting lines adaptively to the distribution of distance data point when detecting a plane by the scan line grouping, and detect a plane very accurately by detecting the plane by the scan line grouping with the user of the lines.

[0056] Also, the above-mentioned plane detection apparatus installed in the robot apparatus according to the present invention permits to detect a plane accurately even if the distance measuring means provided in the robot apparatus has acquired distance data with noises, detect stairs in an environment around the robot apparatus and have the robot apparatus climb the stairs or recognize a step on a floor and have the robot apparatus move on the stepped floor. Thus, the robot apparatus will provide a greater entertainment.

Problems solved by technology

The plane detection unit has to be designed with a smaller threshold for breaking a plane into a plurality of plane segments, which will unavoidably lead to such an under-segmentation.

With the plane detection algorithm using the scan line grouping, however, it is extremely difficult to determine a threshold for the line fitting.

That is to say, with a small threshold, a line will be segmented excessively small under the influence of noises, and it is not possible to extract lines which could normally be extracted.

Generally, data acquired by the distance measuring unit (range finder) such as stereo vision system from a near point is highly accurate and have less noises, but data acquired from a far point is low in accuracy and has many noises.

However, since the variation in measuring accuracy due to the environment will have an influence on the data acquisition, it is difficult to set a unique threshold for such a data acquisition.

In effect, the plane detection based on the randomized Hough transformation is suitable for use to detect a dominant plane, but when applied for detection of a plurality of planes from data including a plurality of planes such as stairs, it will cause the under-segmentation problem.

Also, when the scan line grouping is used to extract planes, it is difficult to set a threshold for the line fitting.

Use of the scan line grouping will cause the over-segmentation and under-segmentation problems.

Both the randomized Hough transformation and scan line grouping are not advantageous in that it is difficult to detect planes accurately from measured distance data.

Images